<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Rds on Vlad Usenko's Blog</title><link>https://blog.vladusenko.io/tags/rds/</link><description>Recent content in Rds on Vlad Usenko's Blog</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Fri, 10 Apr 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://blog.vladusenko.io/tags/rds/index.xml" rel="self" type="application/rss+xml"/><item><title>How Large Queries Broke Our CPU Balance Across Aurora Read Replicas</title><link>https://blog.vladusenko.io/blog/rds-proxy-connection-pinning/</link><pubDate>Fri, 10 Apr 2026 00:00:00 +0000</pubDate><guid>https://blog.vladusenko.io/blog/rds-proxy-connection-pinning/</guid><description>&lt;p&gt;We&amp;rsquo;ve been having troubles with our Aurora MySQL cluster for a while: CPU distribution among read replicas was imbalanced, which led to application performance degradation.&lt;/p&gt;
&lt;p&gt;In this post I will share how we tracked down the root cause, solved it and what RDS Proxy had to do with it.&lt;/p&gt;
&lt;h2 id="the-problem"&gt;The Problem&lt;/h2&gt;
&lt;p&gt;The system in question is a read-heavy service that uses Aurora MySQL as a database with RDS Proxy in between to allow faster failovers and safer maintenance activities.
Traffic is stable most of the time, but there are a few days per month when we expect a surge of read operations to the service, due to which we scale up our Aurora cluster with read replicas preliminarily for a fixed period of time.
However, we&amp;rsquo;ve been seeing an issue with the CPU distribution among the replicas: it was completely uneven, almost random, some nodes were running on almost 99%, some on just a few.&lt;/p&gt;</description></item></channel></rss>